Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!lll-winken!ncis.tis.llnl.gov!blackbird!news
From: news@blackbird.afit.af.mil (News System Account)
Newsgroups: comp.sys.atari.st
Subject: Re: Absoft Fortran
Summary: SAVE command
Message-ID: <1440@blackbird.afit.af.mil>
Date: 7 Dec 89 11:49:24 GMT
References: <891207024528.958022@DMZRZU71-UNI-MAINZ--GERMANY>
Reply-To: dbell@galaxy-43.UUCP (David E. Bell)
Organization: Air Force Institute of Technology; WPAFB, OH
Lines: 37

In article <891207024528.958022@DMZRZU71-UNI-MAINZ--GERMANY> Ritzert@DMZRZU71.BITNET writes:
>                                                 There is a serious bug
>in the compiler (version 2.3):
>
>If You want to use a common block only in some subroutines You have to
>declare it in Your main program. Otherwise You will get unpredictable
>results. This is the only Fortran compiler I have used which behaves
>this way (and I have been writing Fortran programs on many systems).
>
>Michael Ritzert
>mjr@dmzrzu71.bitnet


This is not a bug.  The FORTRAN standard does not require that a common
block that is first initialized in a subroutine be saved after the 
subroutine is exited; although, most FORTRAN implementations do save it.
(Sorry, I can't quote the Standard but you can
see, for instance, Michael Metcalf's Effective Fortran 77, page 83.)

If you would like to keep the common block around and don't want
to clutter up the main program by declaring it there use the save
command:
	   subroutine foo(args)
	   real args,stuff
	   common/bar/stuff
	   SAVE /bar/
	   .
	   .
	   .
This should work (at least it does in compiler version 2.2).

(It sure is nice to finally see some discussion on a
real programming language in this newsgroup :-)

David E. Bell
 dbell@galaxy-43.UUCP
 dbell@afit-ab.arpa